Protein binding site comparison

Abstract

The aim of the present study was the identification of novel hit compounds for the inhibition of trypanothione synthetase (TryS) from organisms of the genus Leishmania. They are the causative agents of the so-called neglected diseases. There is an urgent need for novel therapeutics for these diseases as the current treatment methods are not sufficiently reliable when compared to the medical management for diseases prevalent in developed high-income countries. The basic idea of this work was to set up a structure-based drug repurposing workflow based on similarities between the ATP binding sites of TryS from Leishmania major and protein kinases. The latter enzymes are well characterised, and this approach benefits from the huge knowledge of drug-like small molecule inhibitors of protein kinases. However, the establishment of such a workflow requires the reliable interplay of multiple software methods for in silico structure-based modelling. The chosen workflow includes the comparative modelling of the enzymes under investigation, MD simulation studies, an in-depth analysis of pocket properties, the comparison of binding sites of TryS and kinases, and a subsequent molecular docking of known inhibitors of protein kinases which are most similar to TryS with respect to the ATP binding site properties. The breakneck developments in the field of in silico hit identification led to the emergence of a myriad of available methods to meet similar targets. Current state-of-the-art methods evolved for often applied structure-based methods such as molecular docking or pharmacophore searches. Unfortunately, methods for the prediction, analysis, and comparison of binding sites are less frequently applied. This led to the parallel development of numerous methods which differ in multiple aspects. This diversity necessitates reliable comparison and benchmark analyses to elucidate the most suitable approaches for a pre-defined scientific question. In consequence, a main part of the present work is dedicated to the comprehensive evaluation of developed software for the prediction and comparison of binding sites. The analyses shed light on the major strengths and restrictions of the methods under investigation. The results ensure that only the most reliable methods are chosen for the problem under investigation. For binding site prediction methods, we can pinpoint a restricted set of exceptionally reliable methods. The case is different for binding site comparison tools. We must finally conclude that the establishment of a consensus approach might be the most appropriate way to identify binding site similarities. Besides the comparison of binding sites based on physicochemical and shape properties, an alternative approach was tested for its applicability for the presented project. Based on the observation that some less closely related proteins binding to similar scaffolds share a common arrangement of SSEs in the ligand binding sites, it was hypothesised that finding similar arrangements of SSEs in unrelated proteins might facilitate the search for novel promising scaffold for yet unexplored enzymes. This so-called “ligand-sensing cores” approach was adopted for the comparison of ATP binding sites of TryS and kinases. However, the results for this case study indicate not only that it is difficult to pinpoint specific and unique secondary structure element arrangements in protein binding sites, but also that the success in establishing a novel method rises and falls with the availability of proper optimisation datasets and the reliability of secondary structure assignment methods. We tried to solve at least the latter part of the problem by developing the novel secondary structure assignment tool SCOT which classifies geometrically uniform helices and strands. But especially with respect to parameter optimisation, further developments of SSE comparison methods and the generation of suitable datasets will be necessary. However, with respect to the target of this study, we could show that well-established methods for the comparison of protein binding sites are perfectly suitable to pinpoint similarities between the nucleotide binding sites of TryS and protein kinases. Known inhibitors of protein kinases with the most significant similarities to the target and structurally related compounds were investigated. Finally, we could propose promising molecules for addressing the ATP binding site of TryS which must now be evaluated experimentally for their potential to inhibit TryS from Leishmania.